At the forefront of
technology

“Make difference.”

Mechanical Design

Integrating cooling systems in motors to manage temperature and ensure stability.
Using advanced materials to achieve high torque relative to motor size.
Employing motor drives for precise control of current, voltage, torque, speed, and position, while preventing overload.

Robot Control

Agile Walking Control: Achieved through fast and precise communication between the control system and motor drives.
Quick Balance Recovery: Ensures rapid stabilization in case of impacts or falls.

Natural Walking

Versatile Terrain Navigation: Enabled by deep reinforcement learning for natural walking across various terrains.
Robust Real-World Walking: Achieved through high-fidelity simulators and parallel learning techniques.
Biomimetic Walking Techniques: Developed by applying deep learning-based motion capture of real animals.

AI Perception

Facial Expression Emotion Recognition: Utilizing deep learning models to understand and respond to emotions based on facial expressions.
Gesture and Pose Recognition: Employing deep learning models to interpret gestures and body poses.
Voice-Based Emotion Recognition: Implementing deep learning models to analyze and recognize emotions through voice.

Accurate Situation Recognition and Complex Thinking

Advanced Feature Extraction: Developing techniques for precise feature extraction and feature space projection from vision and voice inputs.
Enhanced Accuracy: Improving performance through explicit and implicit alignment, as well as joint learning methods.

Personality Formation through User Interaction

User Data Collection: Gathering and securing user data to facilitate interactive experiences.
Continuous Learning: Evolving the personality of R.pet through ongoing learning both on the cloud and on devices.

Autonomous Movement and Map Creation

Stable Map Creation: Generating reliable maps based on precise robot location estimates.
Efficient Sensor Data Processing: Rapidly constructing maps by effectively processing data from various sensors.
Noise and Error Minimization: Reducing inaccuracies through the integration of multiple sensors such as RGB cameras and LiDAR.

Motion Planning

Identifying terrain features in various environments, estimating its own location, and selecting optimized paths and actions to reach the destination.
Capable of traversing a wide range of terrains, unlike wheeled robots.
Uses perceived information and map data to plan routes, determining the optimal path and direction.
Adjusts walking speed according to the surrounding environment.
Detects potential collision risks based on perceived information and takes appropriate measures to avoid them.
Maintains a safe distance from moving objects in the vicinity while walking.